INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

chat

-0.78

taboola

-0.77

"]=>

-0.76

irtual

-0.74

Rated

-0.73

monkey

-0.72

redd

-0.69

Wan

-0.69

>>

-0.69

Sport

-0.68

POSITIVE LOGITS

 Corpus

0.73

Fir

0.70

Ney

0.66

ngth

0.64

 whistle

0.64

 Urug

0.64

 Pixie

0.63

SIG

0.63

 Pill

0.62

 Guardians

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.